skills/manaflow-ai/cmux/cmux-testing/Gen Agent Trust Hub

cmux-testing

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Provides specific shell commands for project validation, including xcodebuild for unit and UI testing with specific flags for schemes and derived data paths.
  • [COMMAND_EXECUTION]: Instructs on the use of internal shell scripts for fuzz testing and environment setup, such as scripts/remote-tmux-fuzz-host.sh and scripts/remote-tmux-live-fuzz.sh.
  • [COMMAND_EXECUTION]: Employs SSH for local loopback testing and automation, including commands for SSH host key management (ssh-keygen) and remote command execution over local aliases.
  • [PROMPT_INJECTION]: The skill establishes an indirect prompt injection surface. 1. Ingestion points: project metadata (project.pbxproj) and runtime output (tmux capture-pane). 2. Boundary markers: Absent. 3. Capability inventory: xcodebuild, ssh, and shell script execution. 4. Sanitization: Absent. These ingestion points are standard for the skill's purpose in code validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 09:00 AM
Security Audit — agent-trust-hub — cmux-testing